CHALLENGES IN CANCER CLASSIFICATION

Classification problem has been extensively studied by researchers in the area of statistics, machine learning and databases. Many classification algorithms have been proposed in the past, such as the decision tree methods, the linear discrimination analysis, the bayesian network, etc. For the last few years, researchers have started paying attention to the cancer classification using gene expression. Studies have shown that gene expression changes are related with different types of cancers. Most proposed cancer classification methods are from the statistical and machine learning area, ranging from the old nearest neighbor analysis, to the new support vector machines. There is no single classifier that is superior over the rest. Some of the methods only works well on binary-class problems and not extensible to multi-class problems, while others are more general and flexible. One thing to note for most of those proposed algorithms on gene classification is that the authors are only concerned with the accuracy of the classification and did not pay much attention to the running time(in fact, most gene classifier proposed are quite computationally expensive). Recently, the neural network has become a popular tool in the classification of Cancer Dataset. This is particularly due to its ability to represent the behavior of linear or nonlinear functions multidimensional and complex.